Implementations of wireless active noise control in the headrest

Xiaoyi Shen 1 , Dongyuan Shi, Santi Peksi, Woon-Seng Gan. Digital Signal Processing Lab, Nanyang Technological University 50 Nanyang Avenue, Singapore 639798

ABSTRACT Working from home went from a temporary result of the pandemic to a way of life. However, peace and quiet are hard to come by. To get rid of distraction, Active Noise Control (ANC) particularly the ANC headphones are widely used to attenuate the undesired noise in daily life. While it works, the prolonged usage of headphone can cause ear problems such as wax blockage, ear infections, pain, hearing loss, and impairment. This paper describes a wireless ANC system implementation designed for the headrest to reduce the noise distraction while doing away with discomfort caused by prolonged usage of headphone. Wireless microphones are utilized to pick up the noise with a high reference-to-interference ratio, which helps in improving the noise reduction performance of the ANC headrest. This paper investigates the wireless feedforward, feedback, and wireless hybrid ANC structures’ noise reduction performances in the ANC headrest. Comparative experimental results shown in this paper indicate the e ff ectiveness of the proposed approach in reducing noise around the headrest.

1. INTRODUCTION

Active noise control (ANC) headphone is a a mature device in the market for creating a quiet environment [1–10]. Long-term use of headphones, on the other hand, causes customers to experience discomfort in their heads and ears, as well as a variety of health problems. The ANC headrest is designed to address this issue. The headrest can be used in vehicle and airplane seats, workspaces, and other relaxing environments. Users may relax themselves on the chair and no longer need to wear headphones or earphones. Instead of placing the secondary speaker inside the earcups, the ANC headrest positions it behind the head, as seen in Fig 1. Two secondary speakers and two error microphones are built into the ANC headrest to o ff er quite zones for the left and right ears. The headrests were implemented using a variety of ANC structures. In some cases, the feedforward ANC structures were built into the headrest. During these applications, an infrared sensor that detected the occupant’s head was used in the feedforward ANC system [11], where the filter coe ffi cients were pre-trained for various conditions and selected based on seating position. To improve the tracking performance on the head movement, the infrared rangefinders with the Kalman filter algorithm were brought into the feedforward ANC system [12]. Furthermore, feedback systems were also widely employed in headrests. An adaptive feedback algorithm was used to implement the ANC headrest without reference microphones, and single-channel feedback with combined

1 xiaoyi003@e.ntu.edu.sg, dongyuan.shi@ntu.edu.sg, SPeksi@ntu.edu.sg, EWSGAN@ntu.edu.sg

a slaty. inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS O ¥, ? GLASGOW

Figure 1: The front and side view of the ANC headrest

error signals and combined secondary speaker output was used to replace the two-channel feedback ANC algorithm, saving computational complexity [13]. Moreover, another feedback ANC method with head-mounted error microphones was realized on the headrest [14]. To further improve noise reduction performance, a feedback ANC structure involving a set of adaptive and fixed filters was proposed [15]. According to these studies, we can conclude that a hybrid combination of feedforward and feedback structures will provide numerous benefits to the ANC system and, as a result, is a desirable solution for the ANC headrest. The feedforward, feedback, and hybrid structures used in the ANC headrest in Sec 2 are revisited in this paper. Wireless microphones are used as reference microphones in feedforward and hybrid ANC to obtain a reference signal with a high signal to interference ratio (SIR) [16–19]. Sec 3 evaluates the noise reduction performance of wireless feedforward, feedback, and wireless hybrid ANC in real-time experiments using various noises type (sine-tone noise, broadband noise, real noise, multiple noises). Finally, in Sec 4, a conclusion is made.

2. ANC STRUCTURES FOR HEADREST

This section will provide an overview of the proposed ANC structures and algorithms that will be implemented in the headrest.

2.1. Wireless feedforward ANC structure in the headrest The wireless microphone is used as the reference microphone in the feedforward ANC headrest [20–23]. The block diagram of wireless feedforward ANC is shown in Fig.2. The control signal is generated by the summation of M control filters:

M X

y f ( n ) =

m = 1 y m ( n ) , (1)

“am

Figure 2: The block diagram of wireless feedforward ANC structure.

where y m ( n ) is the output signal of m th control filter w m ( n ):

y m ( n ) = x T m ( n ) w m ( n ) , m = 1 , 2 , · · · , M (2)

where superscript ( · ) T denotes the transpose of the argument. x m ( n ) represents the reference signal picked up by the m th wireless microphone:

x m ( n ) = { x m ( n ) , x m ( n − 1) , · · · , x m ( n − L f + 1) } T . (3)

where L f is the length of feedforward control filter. The filter-x least mean square (FxLMS) algorithm is used to minimize the error signal obtained by the error microphone. The coe ffi cients of m th control filter are updated by: w m ( n + 1) = w m ( n ) − µ f x ′ m ( n ) e ( n ) , (4)

where µ f is the step size of FXLMS algorithm, and the error signal is collected by the error microphone: e ( n ) = d ( n ) + y f ( n ) ∗ s ( n ) . (5)

where ∗ represents the convolution operation. s ( n ) is the impulse response of the secondary path from the secondary speaker to the error microphone, whose estimation ˆ s ( n ) can be modeled either online or o ffl ine, and d ( n ) denotes the disturbance at the error microphone. In practice, the filtered reference signal x ′ m ( n ) in Equation 4 is obtained from:

x ′ m ( n ) = x m ( n ) ∗ ˆ s ( n ) . (6)

2.2. Feedback ANC structure in the headrest From the previous discussion, we can figure out that the noise reduction performance of feedforward ANC is improved by reference signals with a high SIR obtained with wireless microphones.However, if the wireless microphone fails to pick up the noise that becomes the disturbance at the error microphone, the noise reduction performance of the ANC system will su ff er dramatically. Under this situation, the feedback ANC [24–27] would become a good solution since it only relies on one error microphone to acquire the reference and error signal, as shown in Fig 3. The coe ffi cients of feedback control filter W b ( z ) are also updated by using the FXLMS algorithm:

w b ( n + 1) = w b ( n ) − µ b ˆ x ′ b ( n ) e ( n ) , (7)

where the filtered feedback reference is:

ˆ x ′ b ( n ) = ˆ x b ( n ) ∗ ˆ s ( n ) , (8)

The control signal of feedback structure is generated:

y b ( n ) = ˆ x T b ( n ) w b ( n ) . (9)

Noise source 1 M-channel feedforward structure — 7) ) i OT res a | . Sta_iws as) Noise sourceM ) ) — il iii ) Lame ert rs Transmitter oa Error sensor Secondary source A/D e(n)

Figure 3: The block diagram of feedback ANC structure.

d(n) Secondary iii y, (n) source | Error sensor

Figure 4: The block diagram of wireless hybrid ANC structure.

2.3. Wireless hybrid ANC structure in the headrest Although it is easier to implement a feedback ANC headrest because a reference microphone is not required, another factor to consider is the type of noise that can be canceled by the feedback structure. Therefore, a hybrid combination of wireless feedforward and feedback, which possesses the benefits of both feedforward and feedback structures, should be a better choice for the ANC headrest. This hybrid structure [28–30] is shown in Fig 4. The output of the control filter for the feedforward and feedback part is calculated as Equation 2 and Equation 9. The control signal of hybrid structure is generated by the summation:

M X

y ( n ) =

m = 1 y m ( n ) + y b ( n ) , m = 1 , 2 , · · · , M (10)

3. EXPERIMENT RESULTS

Experiments are taken to validate the noise reduction performed by wireless feedforward ANC, feedback ANC and wireless hybrid ANC. A movable chair equipped with a headrest is applied to realize noise reduction. The ANC headrest is shown in front and side views in Fig 1. Two secondary speakers are mounted on the two sides of the headrest to create a quiet zone for the ears. The

Noise source 1 M-channel feedforward structure — 7) ») Brame ee S( e e e e Noise sourceM @ y, (m2) ») pharm fT panel a Lp fae Feedback structure d(n) D/A lo Error sensor Secondary source A/D e(n)

Figure 5: (a) The set up of the ANC headrest, which is worn on a dummy head; (b) The perspective view of the experimental chair equipped with the proposed ANC headrest.

chair’s wheel is designed to make the ANC headrest movable so that noise transmission from various directions is verified. As depicted in Fig 5, a dummy head is utilized to represent the person sitting in the chair. The error microphone (Sound Professionals Ultra-Low-Noise In-Ear Binaural microphones) is placed between the secondary speakers and the dummy head’s ear. Wireless microphones are used for all of the reference microphones, which are set in front of each noise source. In the embedded controller (NI PXIe-8880), the ANC algorithms of wireless feedforward, feedback, and wireless hybrid are implemented. To complete the signal conversion between analog and digital, the analog to digital converter (ADC) and digital to analog converter (DAC) modules (NI PXIe-6368) are used. Before being played through the secondary speakers, the ANC controller’s output is processed to an amplifier. As shown in Fig 5, all of the pieces of equipment (PXI, amplifier, etc.) are installed inside the chair’s cavity. The monitoring microphone inside the dummy head measures the sound pressure level (SPL).

3.1. Broadband noise cancellation The purpose of this experiment is to verify the performance of sine-tone and broadband noise reduction of the headrest. A loudspeaker situated in front of the chair about 1 . 5m distant from the dummy head plays sine tones with frequencies of 300Hz, 500Hz, and 800Hz, as well as the broadband noise with a frequency of 300 − 800Hz. Feedforward and feedback control filters have step sizes of 1 × 10 − 8 and 1 × 10 − 9 , respectively. For wireless feedforward, feedback, and wireless hybrid ANC structures, the filter length is set to 1024 taps. Furthermore, the ADC module’s sampling rates are 16kHz. Table 1 shows noise reduction levels of wireless feedforward, feedback, and wireless hybrid structures. In sine-tones’ attenuation, three ANC structures behaved nearly identically and achieved 18 − 23dB noise reduction level. Among these methods, the wireless hybrid achieved the best noise reduction performance for broadband noise at 22dB, slightly better than the wireless feedforward ANC’s 20dB. In contrast, the feedback structure is the least e ff ective, with a noise reduction of 10dB. Hence, we can find that the feedback ANC structure is unsuitable for broadband noise cancellation. The detailed 1 / 3 octave band of noise reduction and the power spectrum of the attenuated

Noise reduction Wireless feedforward Feedback Wireless hybrid

300Hz 20dB 18dB 20dB

500Hz 21dB 23dB 21dB

800 Hz 19dB 18dB 20dB

300-800Hz 20dB 10dB 22dB

Table 1: The noise reduction levels of wireless feedforward,feedback, and wireless hybrid ANC headrest for sine-tone and broadband noise.

Figure 6: Noise reduction performance of the algorithms with the broadband noise: (a) The 1 / 3 octave band of noise reduction. (b) The power spectrum of the attenuated disturbance.

disturbances for broadband noise are shown in Fig 6. Both the wireless feedforward and wireless hybrid ANC have good noise reduction performance over the frequency band of 300 − 800Hz, as shown in the figure. Although feedback ANC has attenuation between 300 − 800Hz, it has a waterbed e ff ect between800 − 1000Hz.

3.2. Real noise cancellation The noise reduction performance of the ANC headrest is tested using pre-recorded and loudspeaker-played engine and helicopter noise. The loudspeaker is in the same location as the Experiment 1. Table 2 lists the noise reduction levels for these two noises. The engine noise is reduced by 15 − 16dB in both wireless feedforward and wireless hybrid systems, but only by 6dB in the feedback system. However, the noise reduction e ffi cacy of all ANC structures for helicopter noise was only 2 − 3dB. This is due to the fact that helicopter noise is non-stationary, and the traditional FXLMS algorithm cannot handle it. Figure 7 and Figure 8 illustrate the 1 / 3 octave band and power spectrum of attenuated disturbance. The figure clearly depicts the frequency band’s detailed attenuation. For engine noise, wireless feedforward and wireless hybrid ANC successfully reduce noise between 100 − 500 Hz, with the largest noise reduction occurring around 300 Hz. The feedback is likewise e ff ective at frequencies of 100 − 500Hz, but it has a water e ff ect at frequency of 500 − 1000Hz. In the case of helicopter

Noise Reduction(dB) Amplitude (dB) 40 30 20 10 o2) oOo Dp oO i oO nO oO (a) 1/3 Octave Band (Broadband) T T T HEE Wireless feedforward ANC L [EE Feedback ANC J [J Wireless hybrid ANC o 1 DO 315 400 500 630 200 2 800 1000 1250 1600 2000 Frequency(Hz) (b) Power spectrum of attenuated disturbance (Broadband) T T T T T T T T — Without ANC — Wireless feedforward ANC Feedback ANC — Wireless hybrid ANC 0 200 400 600 800 1000 =: 120 1400 1600 1800 2000 Frequency(Hz)

Figure 7: Noise reduction performance of the algorithms with the engine noise: (a) The 1 / 3 octave band of noise reduction. (b) The power spectrum of the attenuated disturbance.

Noise reduction Wireless feedforward Feedback Wireless hybrid

Engine 15dB 6dB 16dB

Helicopter 2 − 3dB 2 − 3dB 2 − 3dB

Table 2: The noise reduction levels of wireless feedforward,feedback, and wireless hybrid ANC headrest for engine and helicopter noise.

noise, the ANC’s poor noise reduction results show no reduction in the noise’s high-frequency region (800 − 2000Hz).

3.3. Multiple noise cancellation The experiment is carried out in the environment with two noise sources in di ff erent places to validate the noise reduction performance for multiple noise sources. The noise source #1 is a broadband noise with a frequency range of 300 − 500Hz, which is located on the chair’s left side. The noise source #2, which has a frequency range of 300 − 800Hz, is played on the right side of the chair. The rest of the settings are the same as Experiment 1. The noise reduction levels of the wireless feedforward, feedback, and wireless hybrid ANC are 10dB, 7dB, and 16dB, respectively. Fig 9 depicts the 1 / 3 octave band and power spectrum of attenuated disturbance. From the result, it can be found that the wireless hybrid ANC suppresses both broadband noises at 300 − 500Hz and 300 − 800Hz. In comparison, the feedback reduces broadband noise at a low level but increases at 800 − 1000Hz. Therefore, the wireless hybrid ANC achieves the best noise reduction performance in this experiment.

3.4. Uncorrelated interference cancellation This experiment is designed to investigate the performance of the proposed method when an uncorrelated noise occurs during the noise reduction processing. It is worth noting that the uncorrelated noise has no coherence with the reference signal. In this experiment, the wireless

Noise Reduction(dB) Amplitude (dB) 40 30 20 10 80 60 40 20 (a) 1/3 Octave Band (Engine) T T T T T T T T T T HEE Wireless feedforward ANC [EN Feedback ANC J [J Wireless hybrid ANC un. dal 200 400 500 630 300 1000 1250 1600 2000 Frequency(Hz) (b) Power spectrum of attenuated disturbance (Engine) T T T T T T T T — Without ANC — Wireless feedforward ANC Feedback ANC — Wireless hybrid ANC 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Frequency(Hz)

Figure 8: Noise reduction performance of the algorithms with the helicopter noise: (a) The 1 / 3 octave band of noise reduction . (b) The power spectrum of the attenuated disturbance.

Noise Reduction(dB) Amplitude (dB) 40 30 20 10 80 60 7 40 20 (a) 1/3 Octave Band (Multiple sources) T T T T T add T T HEE Wireless feedforward ANC [EN Feedback ANC [Wireless hybrid ANC 200 250 315 400 (b) Power spectrum of attenuated disturbance (Multiple sources) T T T T T T 800 1000 1250 1600 2000 Frequency(Hz) — Without ANC — Wireless feedforward ANC Feedback ANC — Wireless hybrid ANC 0 200 400 600 1 800 1000 1200 1400 1600 1800 2000 Frequency(Hz)

Figure 9: Noise reduction performance of the algorithms with multiple noises: (a) The 1 / 3 octave band of noise reduction . (b) The power spectrum of the attenuated disturbance.

microphone in front of the chair picks up a broadband noise with a frequency of 300 − 800Hz, whereas the uncorrelated noise (not picked up by the wireless microphone) with a frequency of 500Hz occurs on the left side of the chair. Other experiment settings are the same as in Experiment 1. The noise reduction levels of the wireless feedforward, feedback, and wireless hybrid ANC methods are 4dB, 8dB, and 17dB, respectively. The power spectrum of attenuated disturbance in Fig 10 clearly demonstrates that the wireless feedforward ANC performs the worst in dealing with uncorrelated interference. The wireless feedforward ANC is unable to attenuate the noise component

Noise Reduction(dB) Amplitude (dB) 40 30 20+ 10+ oe) oOo Dp oOo iN oO NO So (a) 1/3 Octave Band (Helicopter) [EN Feedback ANC HEE Wireless feedforward ANC [J Wireless hybrid ANC 630 800 1000 1250 1600 2000 Trequeney (Hz) (b) Power spectrum of attenuated disturbance(Helicopter) T T T T T T — Without ANC Feedback ANC — Wireless feedforward ANC — Wireless hybrid ANC 0 200 400 600 800 1000 1200 1400 1600 Frequency(Hz) 1800 =2000

Figure 10: Noise reduction performance of the algorithms with the uncorrelated interference: (a) The 1 / 3 octave band of noise reduction . (b) The power spectrum of the attenuated disturbance.

(500Hz) that is not picked up by the wireless microphone. In contrast, the hybrid ANC can deal with both broadband noise (300 − 800Hz) and uncorrelated interference (500Hz) using a combination of the feedforward and feedback structure.

4. CONCLUSION

This paper evaluated the noise reduction performance of wireless feedforward, feedback, and wireless hybrid ANC structures. The use of wireless microphones improved the signal-to-interference ratio of reference signals received. The wireless hybrid ANC performed well whether the noise sources were known (16 − 22dB noise reduction level) or unknown (16dB noise reduction level), whereas the wireless feedforward performed better in environments with known noise sources (10 − 21dB noise reduction level), but dropped to 4dB noise reduction level when the wireless microphone did not pick up the reference signal. Feedback ANC only achieved a noise reduction level of 6 − 10dB when dealing with broadband noise. However, non-stationary noise, such as helicopter noise, is attenuated only by 2 − 3dB noise reduction level by wireless feedforward, feedback, a wireless hybrid structures.

REFERENCES

[1] SJ Elliot and PA Nelson. Active noise control. Noise News International , 2(2):75–98, 1994. [2] Sen M Kuo and Dennis R Morgan. Active noise control systems , volume 4. Wiley, New York, 1996. [3] Dongyuan Shi, Woon-Seng Gan, Bhan Lam, Shulin Wen, and Xiaoyi Shen. Optimal output- constrained active noise control based on inverse adaptive modeling leak factor estimate. IEEE / ACM Transactions on Audio, Speech, and Language Processing , 29:1256–1269, 2021. [4] Bhan Lam, Chuang Shi, Dongyuan Shi, and Woon-Seng Gan. Active control of sound through full-sized open windows. Building and Environment , 141:16–27, 2018.

Noise Reduction(dB) Amplitude (dB) 40 30 20 10 oe) oOo iN oO NO So (a) 1/3 Octave Band (Interference) 7 T T T T T T T T T T HE Wirelessfeedforward ANC [EN Feedback ANC [Wireless hybrid ANC 200 250 = 315 800 1000 1250 1600 2000 Frequency(Hz) Dp oOo (b) Power spectrum of attenuated disturbance (Interference) T T T T T T T T — Without ANC — Wireless feedforward ANC Feedback ANC — Wireless hybrid ANC 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Frequency(Hz)

[5] Dongyuan Shi, Woon-Seng Gan, Bhan Lam, and Shulin Wen. Feedforward selective fixed-filter active noise control: Algorithm and implementation. IEEE / ACM Transactions on Audio, Speech, and Language Processing , 28:1479–1492, 2020. [6] Jordan Cheer and Stephen J Elliott. The design and performance of feedback controllers for the attenuation of road noise in vehicles. International Journal of Acoustics and Vibration , 19(3):155–164, 2014. [7] Bhan Lam, Woon-Seng Gan, DongYuan Shi, Masaharu Nishimura, and Stephen Elliott. Ten questions concerning active noise control in the built environment. Building and Environment , 200:107928, 2021. [8] Dongyuan Shi, Woon-Seng Gan, Bhan Lam, Shulin Wen, and Xiaoyi Shen. Active noise control based on the momentum multichannel normalized filtered-x least mean square algorithm. In INTER-NOISE and NOISE-CON Congress and Conference Proceedings , volume 261, pages 709–719. Institute of Noise Control Engineering, 2020. [9] Woomin Jung, Stephen J Elliott, and Jordan Cheer. Local active control of road noise inside a vehicle. Mechanical Systems and Signal Processing , 121:144–157, 2019. [10] Jordan Cheer. Active control of the acoustic environment in an automobile cabin . PhD thesis, University of Southampton, 2012. [11] Cheng-Yuan Chang, Chia-Tseng Chuang, Sen M Kuo, and Chia-Hao Lin. Multi-functional active noise control system on headrest of airplane seat. Mechanical Systems and Signal Processing , 167:108552, 2022. [12] Rong Han, Ming Wu, Chen Gong, Shangshuai Jia, Tieli Han, Hongling Sun, and Jun Yang. Combination of robust algorithm and head-tracking for a feedforward active headrest. Applied Sciences , 9(9):1760, 2019. [13] Antonius Siswanto, Cheng-Yuan Chang, and Sen M Kuo. Active noise control for headrests. In 2015 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA) , pages 688–692. IEEE, 2015. [14] Jan Buck, Sergej Jukkert, and Delf Sachau. Performance evaluation of an active headrest considering non-stationary broadband disturbances and head movement. The Journal of the Acoustical Society of America , 143(5):2571–2579, 2018. [15] Lifu Wu, Feng Niu, and Yecai Guo. Active noise control pillow based on the combination of the fixed and adaptive feedback structures. Applied Acoustics , 185:108396, 2022. [16] Xiaoyi Shen, Dongyuan Shi, and Woon-Seng Gan. A wireless reference active noise control headphone using coherence based selection technique. In ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pages 7983– 7987, 2021. [17] Xiaoyi Shen, Dongyuan Shi, Santi Peksi, and Woon-Seng Gan. A multi-channel wireless active noise control headphone with coherence-based weight determination algorithm. Journal of Signal Processing Systems , pages 1–9, 2022. [18] Xiaoyi Shen, Dongyuan Shi, Woon-Seng Gan, and Santi Peksi. Implementation of coherence- based-selection multi-channel wireless active noise control in headphone. In INTER-NOISE and NOISE-CON Congress and Conference Proceedings , volume 263, pages 1945–1953. Institute of Noise Control Engineering, 2021. [19] Sheng Shen, Nirupam Roy, Junfeng Guan, Haitham Hassanieh, and Romit Roy Choudhury. Mute: bringing iot to noise cancellation. In Proceedings of the 2018 Conference of the ACM Special Interest Group on Data Communication , pages 282–296, 2018. [20] Feiran Yang, Yin Cao, Ming Wu, Felix Albu, and Jun Yang. Frequency-domain filtered-x lms algorithms for active noise control: A review and new insights. Applied Sciences , 8(11):2313, 2018.

[21] J-D Wu and Mingsian R Bai. Application of feedforward adaptive active-noise control for reducing blade passing noise in centrifugal fans. Journal of Sound and vibration , 239(5):1051– 1062, 2001. [22] Chuang Shi, Rong Xie, Nan Jiang, Huiyong Li, and Yoshinobu Kajikawa. Selective virtual sensing technique for multi-channel feedforward active noise control systems. In ICASSP 2019- 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pages 8489–8493. IEEE, 2019. [23] Hayato Ito, Shoichi Koyama, Natsuki Ueno, and Hiroshi Saruwatari. Feedforward spatial active noise control based on kernel interpolation of sound field. In ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) , pages 511– 515. IEEE, 2019. [24] Piero Rivera Benois, Reinhild Roden, Matthias Blau, and Simon Doclo. Optimization of a fixed virtual sensing feedback anc controller for in-ear headphones with multiple loudspeakers. arXiv preprint arXiv:2110.03586 , 2021. [25] Thomas Schumacher, Hauke Krüger, Marco Jeub, Peter Vary, and Christophe Beaugeant. Active noise control in headsets: A new approach for broadband feedback anc. In 2011 IEEE International conference on acoustics, speech and signal processing (ICASSP) , pages 417–420. IEEE, 2011. [26] Yoshinobu Kajikawa. Feedback active noise control system and its application to mri noise. In 2011 8th International Conference on Information, Communications & Signal Processing , pages 1–5. IEEE, 2011. [27] Yaping Ma, Yegui Xiao, Boyan Huang, Tao Bai, and Xinyu Tan. A robust feedback active noise control system with online secondary-path modeling. IEEE Signal Processing Letters , 2022. [28] Xiaoyi Shen, Woon-Seng Gan, and Dongyuan Shi. Alternative switching hybrid anc. Applied Acoustics , 173:107712, 2021. [29] Piero Rivera Benois, Patrick Nowak, and Udo Zölzer. Hybrid active noise control structures: A short overview. In Speech Communication; 13th ITG-Symposium , pages 1–5. VDE, 2018. [30] Muhammad Tahir Akhtar and Wataru Mitsuhashi. Improving performance of hybrid active noise control systems for uncorrelated narrowband disturbances. IEEE transactions on audio, speech, and language processing , 19(7):2058–2066, 2011.